Enhanced User Experience through Real-time Monitoring: Telefónica Leverages GCP and AI
Intro
e-dialog introduced a solution that analyzes website data in near real time in the Google Cloud and automatically notifies o2 Telefónica when a problem occurs. The approach consists of two pipelines developed in GCP: a training pipeline and a real-time prediction pipeline.
Challenge
Timely detection of outages or issues and subsequent resolution of these problems are critical for smooth website operations and an excellent user experience. The earlier an issue is detected, the faster it can be resolved. Traditional manual methods are too slow to respond to sudden changes in user behavior, which impacts response times and user satisfaction.
Our Google Cloud-based anomaly detection solution has significantly improved our ability to quickly identify and resolve issues, minimizing disruptions and ensuring a smooth, uninterrupted experience for our website users. This proactive approach helps us maintain a high level of reliability and user satisfaction.
Gerald Maier
Senior Analytics Engineer, Telefónica Germany
Measures
The developed training pipeline transforms historical Google Analytics 4 data into BigQuery and trains an ML learning model to determine the expected behavior for various metrics, such as traffic & product page views. The model then continuously analyzes data in near real-time to estimate the likelihood of anomalies. The system alerts stakeholders and provides additional statistics via the Data Studio dashboard (formerly Looker Studio).
Results
The implementation of the Google Cloud monitoring and anomaly detection solution has improved website uptime and reliability by detecting changes in near real-time and immediately notifying the app and web teams. Faster response times have enhanced our ability to proactively address website issues, resulting in higher user satisfaction and reduced potential revenue loss.
- x faster response time
- Clicks + engagement
- % Click-through rate + relevance
